Reading Excerpt Question Lab

Discussion questions written from memory of a book — instead of the passage in front of the group — import plot points, backstories, and spoilers the excerpt never contained. This agent skill creates passage-anchored discussion questions from only the excerpt or summary you supply. Its distinctive approach is an excerpt-locked question lab, not a book-club content generator: every question carries a verbatim evidence phrase from your supplied text, nothing outside the passage is imported, a question may invite interpretation but never presupposes an unsupplied answer, and requests for facts beyond the excerpt return HOLD asking for the relevant passage. No sales, earnings, compliance, or outcome result is promised. What's included: a set of discussion questions, each with its verbatim evidence line, when the passage supports them, an internal source check, and clarification questions under HOLD when the material is missing or contradictory, including the source skill text, clearly labeled sample outputs, and a human-review step. This is a text-based digital product; no AI subscription, external account, or files are supplied with it — you use an AI chat tool of your choice. How it works: Install the included SKILL.md in a compatible assistant, or paste the skill text into a chat assistant, then replace the full-width bracket fields with your own material: ［SUPPLIED_EXCERPT_OR_SUMMARY］, ［GROUP_FOCUS］, and ［SPOILER_BOUNDARY］. The instructions return STATUS: READY with a QUESTIONS: set of discussion questions — each with a verbatim EVIDENCE: line from your passage — plus an internal SOURCE CHECK:, or STATUS: HOLD with a QUESTIONS: set of clarification questions instead. Both statuses use the QUESTIONS: heading; the STATUS: line tells them apart. The status rules and those exact terms are defined in the skill text. Missing essential facts and conflicting sources lead to HOLD, not invented information. The buyer must review and approve outputs. Made for community reading-group facilitators preparing passage-based discussions who already have the underlying facts and want an auditable first draft, not research or automated sending, and designed to be usable from copied inputs without specialist prompt-writing knowledge. FAQ: Does it fill in missing details? No — its instructions prohibit inventing details and require flags for uncertainty; AI output can still be wrong, so you must review it. Do I need a paid add-on, API, or connected account? No — you need access to a compatible AI chat; access limits and any chat-service charges depend on that provider. Is the draft ready to send or publish without review? No — the status and source checks are intended to support human verification first. Will it confirm regulations, market rules, or platform policies? No — it only uses your supplied facts and will flag external verification needs where applicable. Digital instruction text only; no professional service, legal/medical/tax advice, external research, software integration, or independent fact verification is included. The outputs depend on the facts you supply and on your AI model's adherence to instructions. You must check all drafts, applicable rules, consents, and required details before use. Fictional examples are demonstrations, not claims about real customers or performance. No sales, earnings, compliance, or outcome guarantee.